Once you know your way around the Score tab, the next question is what a good setup actually looks like. This guide covers the choices worth making and why they matter, rather than how to make them.

If you need the mechanics first, how to open the Score tab and move, rename, add or remove requirements, see our Evidence Based Scoring Configuration article.

Almost every setup decision comes down to putting a requirement in the right place, and that only makes sense once you know that the five types are scored in genuinely different ways.

The practical consequence is that the same requirement produces a very different result depending on where it sits. Put a genuine capability in Eligibility and it stops being scored, so a candidate with ten years of it gets no more credit than one with six months. Put a hard licence requirement in Preferred Skills and nothing gets flagged when a candidate doesn’t hold it.

An eligibility requirement works best as a pass or fail test built around one of two things. The first, and by far the most common, is a specific credential a candidate either holds or does not: a licence, a professional qualification, a registration, or a certification. “Registered Nurse (NMC),” “Full UK Category C+E driving licence” and “ACCA, CIMA or ACA membership” are all good examples, each names something concrete rather than something open to interpretation.

The second is a minimum amount of a certain kind of experience, where the question is whether a candidate has had that much of it or hasn’t, which is covered further down in this section. Both are genuine pass or fail tests, they’re simply built around different kinds of question, so the one to use is whichever matches what you’re really asking of a candidate.

This is the other kind of eligibility requirement. Broad, general categories of experience are also better handled here than as skills. Required and Preferred Skills work best on specific, measurable capabilities, so something like “hospitality experience” or “customer service experience” is too generic to score precisely as a skill. Where you want to insist on a certain amount of that kind of broad experience, it usually works better as an Eligibility requirement with a minimum years of experience set against it, using the same option covered in our Evidence Based Scoring Configuration article, rather than forcing it into Required Skills.

Wherever you can, pair an eligibility requirement with a matching question on your application form that asks about that exact credential directly, a simple yes/no works best. Tali reads a direct answer to a specific question far more reliably than it can infer the same fact from a CV alone, even when a candidate’s CV makes the same thing obvious to a human reader. If a licence, registration or chartership matters to a role, ask about it directly on the form rather than counting on the CV to mention it in the right words.

A chartership is the clearest example of why. Where something like “Chartered Engineer” or a “qualified accountant” style descriptor appears only in a CV’s personal profile or summary paragraph, and your application form asks no question about it, that credential is missed entirely, even though it would be obvious to anyone reading the CV. The fix is always the same: add a direct question to your form naming that specific credential, so the answer comes back as a plain yes or no.

The same thinking pays off beyond eligibility. For the two or three skills that matter most on the role, a short “give an example of a time you…” question on your application form gives candidates somewhere to describe using that skill in a real situation. That’s what earns Demonstrated evidence rather than a Claim Only result, because a CV that simply lists a skill in a bullet gives Tali nothing to verify it against, whereas a few sentences of context does. One or two of these questions is usually enough, and they tend to tell you more than a longer form would.

That pairing is the ideal setup, but plenty of vacancies only ask for a CV, with no application form questions at all. Where that’s the case, expect eligibility items to come back as not found rather than confirmed either way, because something like a specific licence or membership number isn’t always a detail people volunteer unprompted in a CV. That’s the expected result rather than a sign the analysis has missed something.

What it means in practice is that confirming those items falls to you, so check them with the candidate directly, usually at the point you’re about to progress them or close that action out, rather than assuming an item has been dealt with because nothing has been raised about it. Manually confirming an eligibility item against a candidate is covered in our Evidence Based Scoring: Reviewing a Candidate’s Score article.

Where a role accepts more than one route to the same standard, you can cover all of them in a single eligibility requirement. A finance role that will take “ACCA, CIMA or ACA” is the usual example: name all three acceptable routes in one requirement rather than splitting them into separate ones, and Tali treats the requirement as met when the candidate holds any one of them. That’s the right way to set an either/or requirement up, and it keeps your eligibility list reading the way your advert does. If the requirement also needs to cover an “or equivalent experience” style fallback for candidates who don’t hold one of the named qualifications, that part is worth checking against a few real candidates before you rely on it.

An eligibility requirement with a minimum years of experience only answers one question: does this candidate clear the minimum? It gives no extra credit to the candidate who clears it comfortably.

Take a role that needs at least two years in a particular skill, below which you wouldn’t consider someone at all. Two candidates can both clear that two year minimum and still be a long way apart, one with barely two years and one with ten years of it demonstrated across a full career. If you’d want the stronger of the two to score higher for it, the eligibility requirement on its own won’t do that.

The way to get both effects is to add both. Keep the eligibility requirement for the minimum, and add the same capability separately as a Required or Preferred Skill, so that match strength, experience and recency all continue to count above that minimum. The two aren’t linked automatically just because they cover the same thing, so if you want both, add both items yourself.

Because Tali proposes requirements by reading your advert, a requirement that is purely a must have, with nothing more to assess beyond whether the candidate holds it, can sometimes be proposed twice: once as an eligibility requirement, and once as a separate scored item elsewhere, usually a preferred qualification or a skill. A licence or a professional membership are the typical examples. That happens as a side effect of how Tali reads your advert rather than because anything asked for it to be scored as well.

Where it happens, a candidate picks up scored credit simply for meeting a requirement that was only ever meant to be a yes or no check, which quietly lifts their score for no real reason. It’s worth a look for this in the same pass as the checks below. If a pure must have has also landed somewhere as a scored item, remove the scored copy, unless you’ve deliberately decided you want it scored too for the reason described above.

A few checks sit outside what Evidence Based Scoring can confirm from an application. They’re worth knowing about so you know where the answer comes from instead, and so an unconfirmed item on a candidate reads as expected rather than as something that has gone wrong.

Right to work, DBS and similar legal or compliance checks. Evidence Based Scoring isn’t designed to confirm whether a candidate holds a particular legal or compliance document or clearance, such as right to work or a DBS check, so a requirement built around one shows as unconfirmed whatever a candidate has provided. These continue to go through your normal compliance and post offer process for every candidate, exactly as they did before Evidence Based Scoring, so nothing changes in how you handle them.

Shift patterns, working days and general availability. Something like “available for weekend shifts”, “hybrid, two days on site” or a specific start date isn’t a credential Tali can confirm reliably, even where a candidate answers a matching question directly on the application form. These work better as a plain application form question you read yourself.

The weighting controls how much of the score comes from each part of the assessment. There are three sliders on the Score tab, and by default they’re split:

Those three always total 100% between them. Preferred skills aren’t a fourth share of that total. They’re scored separately, at a relative 20% of wherever you set the Skills slider, so they always contribute something to the score while counting for proportionally less than your required criteria.

The default suits a role where technical capability is the main thing you’re assessing. It’s worth moving away from it when the role isn’t shaped that way.

For a highly technical role, such as a software developer, the usual approach is to keep skills weighted high and behaviours low. The difference between candidates on a role like this usually comes down to technical depth rather than working style, so that’s where the weight should sit. As an indication rather than a rule, a starting point might be Skills at 90% and Behaviours at 10%, adjusted from there once you see how candidates are separating out.

For a people facing role, such as a bar manager, front of house or sales assistant, it’s often worth weighting the other way: shifting weight towards behaviours such as customer focus, adaptability and communication, and reducing the technical skills weighting.

The reasoning is that the technical side of these roles is frequently teachable on the job, whereas someone’s working style generally isn’t. Weighting behaviours higher puts the score where the difference between two candidates actually lies. Again as an indication rather than a rule, a starting point might be closer to Skills 55%, Behaviours 40% and Preferred Qualifications 5%, adjusted for what the role actually needs.

Preferred Qualifications is worth raising above its 5% default where a named certificate or membership genuinely separates otherwise similar candidates, which tends to be in regulated fields where holding the qualification is a real differentiator rather than a nice to have.

Tali proposes requirements by reading your advert, and a few patterns are worth a quick check before you leave the Score tab. None of these take long to fix once you know to look for them.

An advert that lists “excellent communication skills” under a heading like “Essential Requirements” is describing something important, but not necessarily a pass or fail gate. If it’s really a capability you want assessed by strength, it belongs in Required Skills rather than Eligibility, so that match strength, experience and recency can count for the candidate rather than the requirement simply being marked met or not met.

Where your advert names a specific certificate or qualification as a nice to have, it should end up in Preferred Qualifications rather than in Required Skills or Eligibility. Matching the placement to the language you used keeps candidates assessed on the terms the advert actually set.

Tali reads the advert copy for the posting itself, the text in Vacancy Details, sometimes referred to as the Advert Copy. It doesn’t read a separate job description document attached to the vacancy.

That matters in two everyday situations:

Requirements that live only in an attachment. If your advert says something along the lines of “see the attached job description for full details” and the real requirements sit in that attachment, Tali won’t see them and won’t propose them.

This isn’t something you need to rewrite the advert to fix. Once you’ve seen what Tali has extracted, anything missing can be added by hand on the Score tab using ‘Add Skill’, so the practical habit is simply to check the list for gaps and fill in whatever the advert didn’t cover. Adding requirements manually is covered in our Evidence Based Scoring Configuration article.

Shorter adverts reused across postings. Because Tali reads this specific advert rather than a generic template for the job title, two postings with the same title but different advert wording will extract differently. If your team runs a short advert while a fuller job description lives elsewhere, check the advert itself reflects what you want scored on this vacancy.

Where the same role is posted repeatedly, it’s worth setting it up as a vacancy requirement template rather than configuring the Score tab from scratch each time. You can template everything you want for that job, so the right requirements are extracted consistently on every posting instead of depending on how that particular advert happened to be worded.

This is worth doing for any role that comes round regularly: a seasonal hire, a role with high turnover, or a template job title you use across multiple locations. The setup effort goes in once, and every future posting of that role starts from the configuration you already know works.

Setting one up is covered in our Evidence Based Scoring: Vacancy Requirement Templates article.

The feedback option on the Score tab goes back to the product team and is used to improve extraction quality over time, so it’s worth leaving a response rather than skipping past it, particularly on a vacancy where the analysis was noticeably good or noticeably off.

As a quick check before you move on, a vacancy that’s set up well tends to have: